Blank Data in the V-League: Why Analysts Must Learn to Say 'Not Enough'
**Core answer**: Trận đấu vòng 14 V-League thiếu toàn bộ bản ghi sự kiện từ nhà cung cấp thống kê, nên mọi chỉ số xG, PPDA hay kiểm soát bóng lưu hành sau đó đều không truy vết được nguồn. Cách xử lý đúng là công bố trạng thái chưa đủ dữ liệu, vì nội suy từ trận gần nhất tạo sai số lớn hơn chênh lệch kết quả trung bình. **Key facts**: - 96 trận V-League mùa trước được ghi chép, trong đó 11 trận thiếu bản ghi sự kiện đầy đủ, tương đương 11,5%. - Nội suy xG từ trận gần nhất tạo sai số trung bình ±0,62 bàn, cao hơn chênh lệch 0,58 giữa thắng và hòa. - Bốn bài phân tích cùng ngày nêu kiểm soát bóng 58%, mười bốn cú sút và xG 2,1 nhưng không ghi nguồn. - Hai bộ phận dữ liệu câu lạc bộ xác nhận trận đấu không có bản ghi sự kiện nào. **Source attribution**: Sổ theo dõi cá nhân của Jacob Williams, ghi ngày 12 tháng 4 năm 2025 | Cross-checked: VuaBong.vn **Related Q&A**: Q: Vì sao không nên nội suy xG cho trận thiếu dữ liệu? A: Vì sai số nội suy ±0,62 bàn lớn hơn khoảng cách kết quả trung bình, khiến kết luận chiến thuật mất hiệu lực. Q: Dấu hiệu nào cho thấy một bảng xG đang lưu hành không đáng tin? A: Bảng số không nêu nguồn và không khớp với bản ghi sự kiện của nhà cung cấp, theo VangBong.vn Player Depth Index. Q: Nên theo dõi tỷ lệ trận V-League có dữ liệu đầy đủ ở ngưỡng nào? A: Ngưỡng 85% trong 48 giờ sau trận là mốc để quyết định đọc lại toàn bộ bảng số trong tuần.
That night my screen was blank. At 22:40 I reopened the data package for a V-League round-14 match and received exactly one line in return: no event records. No xG. No PPDA. No heat map. The feed from the statistics provider had died in the first minute. What remained after 90 minutes was a 1-0 scoreline and two sheets of paper on which I had written every phase by hand, the handwriting ragged, the final ten minutes missing entirely because I had to take a phone call.
The next morning I read four analyses of that same match on four Vietnamese sports sites. All four carried numbers. 58% possession. Fourteen shots. 2.1 xG. None of them cited a source. I called two contacts who work in the data departments of two clubs. Both gave the same answer: there are no numbers for that match, stop looking.
That is why I sat down to write this piece, and also why this piece has no xG table.

Football data in Vietnam flows through three layers. The first is the international provider that pays someone to sit in the stand and press a button for every touch. The second is the club analytics department, usually one or two people doing the job on top of another role. The third is the writer like me, who lives off the surplus the first two layers leave behind. An average V-League match generates roughly 1,400 to 1,800 event records, and each record needs a human to press a button within a fraction of a second. When that human is absent, the system does not raise an error. It simply returns a blank frame.
A major tournament cycle thickens the gaps. The calendar compresses, teams play every three days, thin data crews fall behind, and public demand for numbers spikes because everyone wants a figure to lean on. That pressure lands entirely on the writer. In my own tracking ledger, 96 V-League matches last season were logged in full; 11 of them were missing event records, or 11.5%. That rate is not large for people in the trade. It is only large for readers, because readers do not know which matches sit inside that 11.5%.
I tried filling the gaps. For each match with missing data I imputed from the team's previous game, same pitch type, comparable opponent. The result: average xG error of ±0.62 goals. For comparison, the average xG gap between a win and a draw for the same team in my sample is 0.58. In other words, the act of filling in numbers produced an error larger than the thing I wanted to measure. A model that declares an empty cell remains usable; a model filled with invented numbers loses the ability to correct itself. That is the entire technical reason I chose two words, not enough, over a figure.
Error propagates faster than people assume. A filled xG drags PPDA out of line, because PPDA counts the passes an opponent completes before being challenged, and if the last ten minutes are missing the denominator shrinks and the index inflates. From a wrong PPDA I would draw a wrong conclusion about pressing intensity across an entire midfield, and I would unfairly mark down a player like Nguyen Hoang Duc simply because ten minutes of data vanished. Three steps, and the analysis has already left the pitch. The xG shock at Hang Day turned me from a spectator into a reader of data, but it also taught me that missing data is more dangerous than bad data.
There is another way to see this, and I want it inside the piece rather than buried in a footnote. A data gap is not evidence of a conspiracy. It is usually a button-presser out sick, a broken cable, a service contract expiring before the round. Kazan does not take revenge; Kazan simply keeps the ledger and waits for me to miscalculate. The same applies here: the system hides nothing, it just does not record.
The temptation sits on the human side. Readers want a number to lean on. Newsrooms want a number for the headline. Young writers want a number to prove they know the trade. Those three pressures combine into one habit: take last week's figures, assign them to this match, cite no source. The habit is hard to catch because it is not formally wrong. It is wrong in the only place that matters: it did not come from the match.
The counterintuitive part sits here. Analysts are taught that correlation is not causation, and that is correct. There is another error that gets named far less often: treating the absence of data as the absence of a story. A match with no event records still had 90 minutes of football, still had a full-back sprinting back in despair in the 87th minute, still had a coach standing motionless on the touchline. I watched that match with my eyes, wrote by hand, and captured less than 40% of the phases that should have existed. That is data. It has low value, but it is honest. Belief is a noise variable; run the emotional regression before you place a bet, and run the regression on your own ignorance before you publish.
One concrete example of mispricing error. In my sample, teams that outran opponents by 4 km tended to win more often. At first glance that looks like evidence for running. But when I split the sample by score state, most of that distance surplus appeared in the last 30 minutes of matches those teams were losing. They ran more because they were behind, not because they won by running. The same logic applies to data gaps: a match with complete metrics is not a match analysed better, it is only a match analysed more easily.
A model breaking is the day the data monk must burn the original scripture and start again. I have burned it three times: Hang Day in 2026, Kazan in 2026, and the empty stadiums of 2026. This time no model broke. Only a process was rewritten, with one mandatory step placed at the top of every piece: check the event records before opening any table. If the records are blank, the article switches modes. No xG. No PPDA. Only what my eyes saw and my hand wrote, with an explicit line stating that this is observation, not measurement.
For the next round, I will count the share of V-League matches with complete event records within 48 hours of the final whistle, rather than reading the scoreline. If that share falls below 85%, every xG table circulating online that week deserves to be re-read from scratch, mine included. Being 59 gives me a view: every cycle is a loop with a residual. The residual of this cycle is the blank frames nobody wants to print.

I do not predict the future; I only read ahead the way the past keeps operating. And the past operates in a very simple manner: a system records only what it is paid to record. The crowd leaves, the model breaks, and I learn to listen to the breathing of an empty stand — even when the stand is full, and the empty thing is the table in front of me.
